DC on output grids

[EFK]

Have (what to me) is a weird issue. This is a new build, Valco type circuit so similar to 1624T with a paraphase feeding a pair of 6V6. On start up, I am getting about .1 vdc on the output grids, but as the amp warms up and runs, it slowly begins to climb and as it does so of course it starts to hum louder and louder, while meanwhile the tubes get hotter, plate voltage drops and cathode voltage climbs. It is climbing up to @ 2-3 vdc on each output grid 1. Say over the course of about 5-10 minutes.

I have swapped output tubes, no change. I have replaced both paraphase caps, thinking maybe a bad one (both new), but no change. I have checked resistances on all the paraphase resistors both cold and after running, no difference, and I've checked out the cathode resistor (Cathode biased output) and there is no noticeable change either cold or hot. All grounding checks out. Output sockets are new and I can't find any problems with either of them. Also, it will do this if I yank the paraphrase tube and run the amp that way with no paraphase (12AX7) in place. I feel like I'm missing something simple but I just can't find it! Where is this DC coming from? Any ideas appreciated.

No cathode bypass on output, but even if I tack one on, it makes no difference.

[g1]

Can you post the 'near' schematic? How do the grids get their ground reference?

[EFK]

Ground reference is back through the paraphase: 270K/4.7K for inverting phase (so say 275K), 270K for non inverting phase. If I read the resistance to each grid right off each output socket all is correct whether amp is cold or has warmed up. Just wondering if there is something I'm missing or not considering here. This is a first for me - I'm no genius but usually I can figure out this simple stuff.

Shared cathode paraphase, voltages are exactly where I would expect them to be, both plate and cathode. They do not change while this is happening. I'm almost wondering if I got TWO bad sets of output tubes?

[mozz]

It does sound like the tubes are gassy or leaking.

[EFK]

Well ok this was with two new sets of Tung Sol 6V6 (the Russian stuff, not old Tung Sol). So I tried it out with a set of nos RCA tubes, and there is indeed a change. I get around 40 mV of DC on the grids at start up and then very slowly it begins rising to @ .1 to .2 vdc after running a bit. Much lower. I don't hear any increasing hum and since I probably would not have even been measuring this in a new build if it wasn't for the increasing hum, I'm wondering now if here is a "normal" range of dc voltage to be found on the output grids? I'm going to have to measure some other amps to see. I am still bugged by the slow increase even with the RCA tubes, but again, I have no idea if this is normal or not.

Despite all the commentary to be found online about how crappy new tubes are supposed to be, I must have been (previously) blessed because I usually do not have any problems with them. I find it hard to believe that two new sets of allegedly-tested or "burned in" tubes would create such a problem.

[Leo_Gnardo]

I find it hard to believe that two new sets of allegedly-tested or "burned in" tubes would create such a problem.

Depends on who did the testing or "burning in" and how honest/thorough they were about it. Five minutes or less in a pinball machine style tester tells you not very much: the tube has emission and isn't shorted out, that's about it. 24 hour burn in? 100 hours? If the tube hasn't flat out failed, sell it.

IIRC, if material from the cathode's coating gets onto the control grid, via sloppy handling during manufacture, that can cause the grid itself to start emitting electrons as it heats up = bad news. A couple tenths of a volt drift, no biggie. A couple volts? Rotten tube, fit for the rubbish tip.

[mozz]

I do have at least 1 Russian tungsol 6v6 that runs away. I had a similar problem but with 6ar6 (5881 clone sshh) cathode biased that cathode voltage slowly crept up. Don't remember if I lowered the grid resistor or it was oscillating, will dig it out and see what I did. Does the filament have a ground reference? I also remember on the tube tester it would slowly show leakage if I left it on a long time.

[Chuck H]

I've had THREE new sets of EL34's fail consecutively in a new build that had 475Vp. Biased at around 75% and screens fed through a 1k resistor rather than a choke. I was starting to think there was something wrong with my build because IT was the common denominator. Nope. As it happens 475V is too much for these Rusky tubes that provide data sheets which simply vomit out the copied, vintage spec of 800Vp max. Don't believe it. The data sheets are cartoons. That is, they look vaguely like real life but if you try that stuff (hold my beer) you're going to get hurt.

Still... Since you also have a vintage set of tubes (old or not) doing the exact same thing I'm inclined to think it may not be the tubes. I'm wondering about your layout, how close the tubes are to each other and lead dress. I'm thinking the amp is oscillating at some very high frequency. You can test for this by clipping (or soldering) something like a 1n cap from each power tube grid to ground and see if the problem persists.

[pdf64]

Ground reference is back through the paraphase: 270K/4.7K for inverting phase (so say 275K), 270K for non inverting phase. If I read the resistance to each grid right off each output socket all is correct whether amp is cold or has warmed up...

Is that including the grid stopper? ie the measurement needs to be made from g1 socket terminal to 0V.
If there is a stopper, maybe the wrong value was fitted, eg way too high/low.
If no stopper, then maybe try fitting them.

[jmaf]

Disconnect the paraphase input and test just the output tubes on their own.

You mentioned that the ground reference is back through the paraphase via 270k resistors, this doesn't make sense to me since you need coupling capacitors between the paraphase outputs and your 6V6 grids. They're 2 different circuits.

The 6V6's need to run fine on their own before you connect the PI to them, including grid leak and the self biasing system. The 6V6 ground reference must come from itself, not from the paraphase resistors.

[EFK]

I think of the paraphase a bit differently because of the way the second grid of the paraphase derives it's signal, so maybe I'm describing it incorrectly. Following the cap output off the first plate (inverting), there is a voltage divider that feeds the input to the non-inverting stage of the paraphase - 270K and 4.7K. This also serves as the ground reference for the first 6V6. Then of course the cap off the second stage uses just a straight 270K to ground which is the ground reference for the second 6V6.

The RCA tubes have settled down after running for a few hours and I'm only getting @ 40mV of DC on the grids (40 on one, @ 30-35 on the other). However, good point about oscillation - I've not had such a problem with this circuit previously, but there are no grid stoppers so maybe I'll tack a couple on there and see if it makes a difference. Up to around 5K or so I can't see how they'd make any audible difference.

I don't think it's running the 6V6 excessively hard, at least not compared to original amps etc. Plates are around 375, screens at @ 360 (shared 2.2K dropping R), idling around 12-13W. Maybe this is just too much for the new Tung Sol tubes? It's actually pretty close to a tweed Deluxe output (aside from the screen R value/screen voltage) with most of the common power transformers people use for those.

I'm going to try out some 6CZ5/6973 with adapters too, and see what those tubes will do.

[jmaf]

I think of the paraphase a bit differently because of the way the second grid of the paraphase derives it's signal, so maybe I'm describing it incorrectly. Following the cap output off the first plate (inverting), there is a voltage divider that feeds the input to the non-inverting stage of the paraphase - 270K and 4.7K. This also serves as the ground reference for the first 6V6. Then of course the cap off the second stage uses just a straight 270K to ground which is the ground reference for the second 6V6.

Could you please draw this for us? I think it'll become clearer once we can look at a schematic.

From what I understand of your description the DC offset would be coming from the PI.

[g1]

I think this is the PI arrangement: http://www.tdpri.com/attachments/162...to-jpg.183779/

[Rhodesplyr]

FWIW, I have observed new production Tung Sol 6550s to exhibit an odd quirk in Leslie amps. They will develop a DC voltage on the grids within around 10 minutes of first being used, but if you let them burn in for an hour or so, it usually goes away and stabilizes. Another Leslie tech pointed this out to me, and I confirmed.

But you'd have to monitor this to tell the difference between an incurable grid-emission problem that could lead to thermal runaway and a getter that needs time to neutralize the remaining gas in the tube from production.

Remember that where you buy tubes is very important. Some places do a burn-in and some don't. Some will say they do, but don't.

[EFK]

G1: that is the circuit exactly, although I am using the first version with the 6SQ7 as the trem oscillator and 6V6 in the output. Other than that, all the same.

The RCA tubes have stabilized with no more than like 10 to 15 mV on the output grids, after running for an hour or so at a time (multiples) over a couple of days. I popped some 2.2K stoppers on the sockets since I had the blank pins anyway but it made no difference at all.

Rhodesplayer I wonder if that is the case here since both new pairs of tubes are doing it, one less than the other but still both doing it. I've not let either set run for more than around 10-15 minutes or so however so maybe I should just run a set for a couple of hours and see what happens. Might give that a try.